IRS-Enhanced Parallel Computing and Partial Offloading for Latency Sensitive MEC

In the rapidly evolving landscape of wireless communications, the integration of intelligent reflecting surfaces (IRS) and mobile edge computing (MEC) has emerged as a transformative paradigm for mobile computing. With the proliferation of data-intensive applications, improving latency purely from a communication perspective no longer suffices to meet the demand on ultra low latency. Unlike prior efforts focusing on IRS-enhanced communication, this letter harnesses the IRSs’ beam manipulation capabilities to improve computation efficiency, where the task of each user equipment (UE) is divided and executed both locally and at multiple edge servers concurrently. Through the joint optimization of IRS phase shifts, UE-IRS associations, offloading ratio, and task distribution ratio, the IRS-enhanced MEC system aims at minimizing maximum delay. Since the optimization problem is non-convex and NP-hard, the problem is decomposed into two subproblems and solved using an efficient block coordinate descent-based algorithm employing the penalty method, successive convex approximation, and reformulation linearization technique. The proposed algorithm converges quickly, with reduced delay up to 49.14% compared to the existing IRS-aided communication scheme.